How can you compare a carrier’s on-time performance by lane instead of relying on one overall score?
Key Takeaways
You compare a carrier's on-time performance by lane by measuring on-time pickup and on-time delivery for each carrier on each lane, from your own load records, using the same definitions for every carrier. A single overall score hides the lanes where a carrier is strong and the lanes where it struggles. A lane view shows which carrier to offer the next load to. An AI logistics agent such as Wilson, by Cartage, records the confirmed pickup and delivery times for every load it handles and can produce carrier-performance reports by lane on request.
- One overall score mixes lanes, modes and customers that have nothing in common.
- Define on-time before measuring: against the appointment, the window or the requested date.
- Measure pickup and delivery separately. They fail for different reasons.
- Use confirmed times with a known source, not estimates or carrier self-reports alone.
- Count enough loads to be fair, and show the count next to the percentage.
- Separate carrier-caused lateness from shipper- or receiver-caused delays.
Why is one overall on-time score misleading?
One overall on-time score is misleading because it averages lanes that behave very differently. An AI logistics agent can break the same data down by lane, which is where tender decisions are made. A carrier's average tells you little about the next load on a specific lane.
| Carrier | Overall on-time | Lane A: Ohio to Georgia | Lane B: Texas to California | Lane C: local drayage |
|---|---|---|---|---|
| Carrier 1 | 91% | 98% (60 loads) | 72% (20 loads) | Not used |
| Carrier 2 | 88% | 85% (30 loads) | 94% (35 loads) | Not used |
| Carrier 3 | 93% | Not used | Not used | 93% (45 loads) |
The numbers are hypothetical. By overall score, Carrier 1 looks better than Carrier 2. On Lane B, Carrier 2 is far more reliable. If the next load is on Lane B, the overall score points to the wrong carrier.
What does a lane mean for this comparison?
A lane is a repeating origin and destination pair, sometimes grouped into regions. For truckload, it may be plant to customer city. For LTL, where shipments go to many destinations, a lane may be origin plant to destination region or state. The grouping should be fine enough that the lanes behave alike, and broad enough that each has enough loads to compare. Many teams start with their top lanes by volume and group the rest by region.
Where does the overall score come from?
The overall score usually comes from a carrier scorecard or a monthly report that adds up every load a carrier moved. It is easy to produce and easy to read. It is useful for a high-level review of a carrier relationship. It is not useful for choosing the carrier for a specific load.
What decisions depend on the lane view?
Several decisions depend on the lane view: which carrier gets the first offer on a lane, which quote wins when prices are close, which carriers stay approved as backups, how much of a service premium is worth paying on a lane with strict customer appointments, and which carriers to invite when the lane is rebid. Each of these is a lane decision. An overall score gives the same answer for every lane, which is why it so often gives the wrong one.
How does lane performance relate to customers?
Lane performance relates to customers because each lane usually serves specific customers with specific requirements. A lane to a big-box retailer with strict appointments and late fees needs a different level of reliability than a lane to the company's own warehouse. Looking at on-time by lane, and by customer within the lane when needed, ties carrier choice to the commitments the company actually made.
Why do teams still rely on one score?
Teams rely on one score because lane-level data is hard to collect by hand. Confirmed pickup and delivery times sit in emails, carrier portals and dock logs. Pulling them together per lane takes hours each month, so the team settles for the carrier's own report or a rough overall number. When the times are recorded on every load as part of the work, the lane view costs nothing extra to produce.
Why do carriers perform differently by lane?
Carriers perform differently by lane because their networks, drivers and customers differ by region. A carrier may have a terminal and steady drivers in one region and rely on interline partners or one-off trucks in another. Traffic, weather, border crossings and dock practices at specific receivers also vary by lane. A lane view captures these differences.
How should on-time be defined?
On-time should be defined the same way for every carrier, against a clear reference, and separately for pickup and delivery. An AI logistics agent can apply the definition automatically to every load. Without a shared definition, comparisons are not fair.
| Measure | Reference | On-time if |
|---|---|---|
| Pickup | Pickup appointment or window on the rate confirmation | Departure confirmed within the window |
| Delivery with appointment | Delivery appointment | Arrival at or before the appointment |
| Delivery with window | Delivery window | Arrival within the window |
| Delivery without appointment | Requested delivery date | Delivered on or before the date |
Who should agree on the definition?
The logistics lead should set the definition with input from customer service and, for key lanes, sales. Customer service knows which customers care about exact times. Sales knows which commitments were made. Once agreed, the definition should be shared with carriers too, so they know how they are measured.
Should pickup and delivery be measured separately?
Yes. Pickup and delivery fail for different reasons. A late pickup is usually about the carrier's truck availability. A late delivery can come from a late pickup, transit problems or receiver delays. A carrier can be reliable at delivery once it picks up, but often late to pick up. Measuring them separately shows where the problem is.
What about loads that were rescheduled?
Loads that were rescheduled with the company's agreement should be measured against the new time, not the original. If the shipper moved the pickup to the afternoon, a carrier that arrived in the afternoon was on time. If the carrier asked to move the pickup and the company accepted, many teams still count it as on time against the new window but track the request separately. The rule should be written down and applied the same way to every carrier.
Should early count as on-time?
It depends on the receiver. Many receivers accept early deliveries, and early arrival counts as on-time. Others refuse early trucks or charge for them. The definition should match the receiver's rules. For appointment-based receivers, arriving well before the appointment can be a problem too.
Should the definition differ by mode?
The core definition should be the same, but the reference points differ by mode. For truckload, the pickup window and delivery appointment on the rate confirmation are the references. For LTL, the pickup date and the carrier's published or quoted transit time are common references, since LTL delivery dates are often estimates. For drayage, the port pickup and the delivery appointment matter. For parcel, the carrier's service commitment is the reference. Each mode's reference should be written down so comparisons within the mode are fair.
What is the right tolerance?
The right tolerance is a business choice: some companies count a pickup within fifteen minutes of the window as on-time; others require it inside the window. The tolerance should be the same for every carrier and written down. Changing it changes every carrier's numbers, so it should not change often.
Where should the data come from?
The data should come from confirmed times with a known source, recorded on each load. An AI logistics agent records these times as part of tracking. Estimates, carrier self-reports and memories are not reliable enough for fair comparisons.
Why do carriers without an API matter here?
Carriers without an API matter because they are often the regional specialists that perform best on certain lanes. If the data only comes from API-connected carriers, those specialists vanish from the comparison, and the lane view is biased toward large carriers. Wilson tracks every load with the best source available, including email, carrier websites and calls, so non-API carriers are measured on the same terms.
Which sources are reliable?
The most reliable sources are the shipping site's confirmed departure or the signed BOL for pickup, and the receiver's check-in or the signed POD for delivery. Carrier tracking events and GPS or ELD data are useful when the source is recorded. Wilson asks the warehouse or receiver for check-in and departure times where configured, records each arrival claim with its source and compares it with the site's record. It marks a load picked up only on a signed BOL or a confirmed departure.
| Source | Reliability for on-time | Notes |
|---|---|---|
| Shipping site departure or signed BOL | High | Firm evidence of pickup |
| Receiver check-in or signed POD time | High | Firm evidence of delivery |
| Carrier tracking events | Medium to high | Depends on the carrier's process |
| GPS or ELD data | High for location | Arrival at the site is not always check-in |
| Carrier email or call | Medium | Useful when recorded with time and source |
| Coordinator's memory | Low | Not usable for comparisons |
Why do carrier self-reports need checking?
Carrier self-reports need checking because carriers report from their own systems, which may record a dispatch time, a GPS arrival near the site or a driver's entry. These can differ from when the truck was actually accepted at the dock. Self-reports are still useful, especially when no site record exists, but they should be labeled with their source. When both exist, the site record usually decides.
What if sources disagree?
If sources disagree, the comparison should use the source closest to the event, and the disagreement should be recorded. For example, a carrier may report arrival at 7:55 while the receiver's check-in shows 8:40. The receiver's check-in is usually the better reference for delivery, because it is when the receiver accepted the truck. The disagreement itself is useful: it may support or contradict a detention charge.
How should missing times be handled?
Missing times should be followed up while the load is fresh, not reconstructed later. If the receiver's check-in time is missing, the agent can ask the receiver or the carrier shortly after delivery. If it remains missing, the load should be marked as "time not confirmed" rather than guessed, and left out of the percentage with a count of excluded loads shown. That keeps the comparison honest.
Why must every load be included?
Every load must be included because leaving some out biases the numbers. If only loads with problems get recorded, every carrier looks worse than it is. If only loads with tracking get recorded, carriers without APIs disappear from the comparison. An AI logistics agent records the times for every load it handles, so the comparison covers the full picture.
How should the comparison be built?
The comparison should be built per carrier per lane, with the load count, on-time pickup, on-time delivery and the main causes of lateness. An AI logistics agent like Wilson can produce this as a carrier-performance report on request from the loads it has handled. Wilson does not keep a standing scorecard by default.
What should the report show?
The report should show, for each lane and carrier:
- Number of loads in the period.
- On-time pickup percentage.
- On-time delivery percentage.
- Average minutes late when late.
- Main causes of lateness, with the share caused by the carrier.
- Cancellations and declined tenders on that lane.
Should recent loads count more?
Recent loads often matter more, because carriers change: a new terminal, a lost customer, a change in drivers. Some teams look at the last quarter for decisions and the last year for context. Others show both side by side. A carrier whose last month is much worse than its year may have a new problem worth asking about.
How many loads are enough?
Enough loads means enough that one bad week does not decide the result. There is no single number for every lane. A carrier with two loads on a lane, one late, shows 50% on-time, which says little. Showing the load count next to every percentage lets the reader judge. For lanes with few loads, a longer period, such as a quarter, gives a fairer view.
How should multi-stop loads be measured?
Multi-stop loads should be measured at each stop, because each stop is its own commitment. A truck that arrives on time at the first stop and late at the third has one late delivery, and the cause may be the first stop's dock. Recording arrival and departure at each stop lets the comparison show where the time was lost.
Should lateness be split by cause?
Yes. Lateness caused by the carrier, such as no truck or a driver out of hours, should be separated from lateness caused by the shipper or receiver, such as freight not ready or a dock that held the truck. A carrier should not be marked down for waiting at a dock it could not unload at. The cause should be recorded on the load at the time, from the carrier's explanation and the site records.
| Cause | Counts against carrier? |
|---|---|
| No truck, late dispatch, driver hours | Yes |
| Breakdown | Usually yes |
| Freight not ready at pickup | No |
| Receiver dock delay or refused appointment | No |
| Weather or road closure | Case by case |
| Border or customs hold | Case by case |
How should transit time be included?
Transit time can be included as a separate measure: how long the carrier took from confirmed pickup to confirmed delivery, compared with its quoted transit time. This separates carriers that pick up late but deliver on schedule from carriers that pick up on time but take longer in transit. Both affect delivery, but they call for different conversations with the carrier.
What about cancellations and declines?
Cancellations and declined tenders should be shown alongside on-time performance. A carrier that is always on time but declines half the loads it is offered on a lane is not a reliable primary carrier for that lane. Acceptance rate and cancellation rate complete the picture.
How should the lane view be used?
The lane view should be used to set the order of carriers on each lane, to decide tie-breaks between quotes and to start conversations with carriers. An AI logistics agent applies the resulting rules on every load. The data informs the team's decisions; it does not replace them.
How should price and performance be balanced?
Price and performance should be balanced by a written lane rule, not case by case. The rule can say how much more the company will pay for a more reliable carrier on that lane, as a percentage or an amount. On lanes with strict customer appointments, the premium may be higher. On lanes to the company's own warehouses, it may be zero. The lane view supplies the reliability; the rule decides how much it is worth.
How does it change the next tender?
It changes the next tender by deciding which carrier gets the first offer and which quote wins when prices are close. A lane rule might say: offer first to the carrier with the best on-time delivery on this lane over the last quarter, if its rate is within the limit; if two quotes are within a set percentage, choose the more reliable carrier. For more on using these results in quote decisions, see how to compare freight quotes from approved carriers.
What about lanes with only one carrier?
On lanes with only one carrier, the lane view still matters. It shows whether that carrier is reliable enough, and it gives the team facts for a conversation with the carrier or for a decision to approve a second carrier. A lane where the only carrier is often late is a lane at risk, and the data makes that visible before a key customer complains.
How should the results be shared inside the company?
The results should be shared with the people who make or feel the decisions: the logistics team for tendering, customer service for setting expectations, and sales for key accounts. A short lane summary each month or quarter is usually enough. When customer service knows that a lane runs reliably, it can give customers firmer delivery promises.
How should carriers be told?
Carriers should be told about their lane performance with the facts: the loads, the dates, the times and the causes. Many carriers improve quickly when they see a clear pattern, such as late Monday pickups on one lane. The conversation is more productive with lane data than with a single overall score the carrier will dispute.
What should happen when a carrier improves?
When a carrier improves on a lane, the next refresh should show it, and the lane order can change. Carriers that respond to feedback should see the result in more loads. This makes the conversation with carriers fair: the data shows problems, and it also shows improvement.
How often should the view be refreshed?
The view should be refreshed on a regular schedule, such as monthly or quarterly, and when something changes on a lane, such as a new customer or a new receiving site. Requesting a report at each review keeps decisions based on recent data.
How does the lane view help in a rebid?
The lane view helps in a rebid because it shows which carriers delivered on each lane, at what rates and with what reliability. When the company asks carriers to bid on lanes for the next period, it can invite the ones that performed, weigh their bids against their record and explain to others what they need to improve. Wilson can produce lane-rate reports on request alongside carrier-performance reports for this purpose.
What should the lane view not be used for?
The lane view should not be used to punish a carrier for a single bad week, or to compare carriers on lanes with very few loads. It is a tool for steady decisions, not for reacting to every incident. For a broader look at this idea, see why the best carrier for the next load depends on the lane.
How does Wilson, by Cartage, support lane-level performance?
Wilson, by Cartage, is an AI logistics agent that books and tracks loads, records confirmed pickup and delivery times with their source, records the cause when a load is late and keeps every load on one record. Because Wilson does the tracking as part of moving each load, the data for a lane comparison is collected along the way, including loads moved by carriers without an API. Wilson does not keep a standing carrier scorecard by default, but it can produce carrier-performance and lane-rate reports on request from the loads it has handled.
Wilson replaces the TMS for freight execution and connects to the company's ERP. It handles LTL, FTL, ocean, drayage, parcel and cross-border freight and books only carriers the company allows.
What does a lane report look like?
A hypothetical lane report for one quarter on a lane from Ohio to Georgia might show three carriers. Carrier 1: 48 loads, 96% on-time pickup, 94% on-time delivery, two late deliveries caused by the receiver's dock. Carrier 2: 15 loads, 87% on-time pickup, 80% on-time delivery, mostly Monday pickups. Carrier 3: 4 loads, all on time, but too few to judge. The team keeps Carrier 1 as primary, asks Carrier 2 about Mondays and offers Carrier 3 more loads to build a fair record.
What stays with the team?
The team defines on-time, sets the tolerance, decides how lane results affect the carrier order and talks to carriers about their results. Wilson applies the lane rules the team sets.
How does Wilson use lane rules when booking?
When the team sets lane rules that use performance, such as a preferred primary carrier or a tie-break for reliability, Wilson applies them on every load: it offers the load in the set order, compares quotes under the rule and books within the limits. The rule in force is recorded with each booking.
How is Wilson priced?
Wilson is free on Wilson's network: the shipper pays the freight rate and nothing on top. When a shipper keeps its own carriers, the fee is 2% to 0.5% of what Wilson pays them, falling with volume. There are no seat fees, no setup cost and no contract.
Can past loads be included?
Loads that Wilson handled are recorded with their confirmed times. Loads from before Wilson usually lack consistent times and sources, so they are best kept separate or left out of the comparison. The lane view becomes reliable as Wilson's records build up.
How do teams start?
Most teams start Wilson on a few lanes and expand from there. Cartage helps map the highest-impact workflow during discovery.
FAQs
How can you compare a carrier's on-time performance by lane instead of relying on one overall score?
Measure on-time pickup and delivery per carrier per lane from confirmed load times, using one definition for all carriers, and show the load count with each result. An AI logistics agent like Wilson, by Cartage, records these times on every load and can produce lane reports on request.
Why is an overall score not enough?
An overall score averages lanes with very different conditions. An AI logistics agent can break the same loads down by lane, which shows the carrier to choose for the next load on that lane.
Should late deliveries caused by the receiver count against the carrier?
No. An AI logistics agent records the cause of each late load, so lateness from receiver dock delays or freight not ready can be separated from carrier-caused lateness.
Does Wilson keep a carrier scorecard?
Not as a standing feature by default. Wilson, an AI logistics agent, records the data on every load and can produce carrier-performance and lane-rate reports on request.
Sources
- Cartage, Wilson product FAQ and product scope, November 2026: https://cartage.ai
If your carrier reviews still rely on one number per carrier, book a demo to see how Wilson, the AI logistics agent by Cartage, records the lane-level facts behind every load.